Thrombocytopenia is extremely frequent in critically ill patients. However, the role of acute platelet responses in critically ill patients is not well studied, and the multifactorial etiology of thrombocytopenia in the ICU makes it difficult to understand, or understand whether or not to treat it. In several situations such as traumatic injury or sepsis, very low platelet counts have been to bleeding, thrombosis and end-organ injury. Platelets have been extensively studied as a key component of hemostasis, but a rapidly emerging concept is that platelets are also key effector cells in systemic inflammatory processes as both instigators of local and systemic inflammatory reactions and also participants in the inflammation that contributes to tissue injury. The link between platelets and inflammation is complex and bidirectional, as inflammatory ligands have been shown to regulate platelet function and activated platelets induce inflammatory responses in other cell types. The overarching theme of this proposal is to study platelet dynamics in critically ill patients, construct clinical endotypes of thrombocytopenia in this population, and to relate these endotypes to underlying mesoscale mechanisms through computational modeling. We will use a large electronic health record-based database and a tri-state trauma database as source data to construct these endotypes. We define endotype as clinical patterns defined along four dimensions: (1) baseline information (demographic, chronic disease burden, severity of illness and admitting diagnosis), (2) features of the platelet count time series (rate of decrease, nadir, etc.), (3) concurrent interventions, and (4) outcome. The computational approach will attempt to root clinical endotypes in mechanistic interpretations (or collections of alternative interpretations), contributing to focus basic science investiagtions, and to close key knowledge gaps preventing the design and use of targeted anti-platelet-inflammatory therapies in the critically ill. Computational models will be developed at different levels of complexity, with a specific attention to tie underlying mechanisms to functional assays routinely performed in thrombocytopenic patients, such as prothrombin time, activated coagulation time, and thromboelastogram.